Tactile based fuzzy-Bayesian controller for prosthetic hand

Abstract

In the experience documented in this paper, we have demonstrated the ability of our novel fuzzy-Bayesian approach to machine cognition, called fuzzy-Bayesian inference system, by making it control a virtual prosthetic hand according to the feedback from its tactile sensors. The hand model is designed based on the characteristics of an average adult hand and is simulated using Java 3D technology with our own developed precise collision determination algorithm and tactile sensors simulator. In this experience, this hand could successfully learn how to pick a fragile subject like a glass without hurting it, in a few experiments.